Chapter 4 illustrates and discusses in detail the steps taken to evaluate the research questions. It first clarifies the methodological approach and then introduces the study design. The declared aim is to create a study that combines the larger number of tokens per speaker created by an experimental framework with the sample diversity and natural environment needed for sociolinguistic analyses. Further sections deal with the sample and the procedure. In this study, seventy-nine speakers with various gender, ethnicity, and age profiles were asked to complete naming tasks, a reading task, a re-narration and a free interview with a meta commentary section.

Chapter 5 describes the patterns found in the data set with a focus on /str/-clusters. Theanalysis starts with a general description of data dispersion and the nature of sibilants in the data set. It addresses the distribution of the data, showing that /str/-clusters are more likely to be retracted than pre-vocalic realizations of /s/. This section further establishes how the distance between /s/ and /ʃ/ can be described by center-of-gravity measurements. Following this operationalization, the focus is on internal factors of language change, namely the question of phonetic and lexical gradualness. Both types of gradualness are confirmed in the analysis of the sibilant space.

The following section focuses on a descriptive statistical account of further social factors in the description of /str/-lowering, of which ethnicity and age are the most relevant. Together with the linguistic factors, the final mixed-effects regression model is then developed and fully evaluated, indicating sibilant duration, age, and ethnicity as most important factors in the change. These results are further confirmed by a random forest regression.

In a qualitative extension of the insights gained through the regression analysis, a brief thematic analysis summarizes the patterns found in the commentary part of the interviews.